truecourse-ai Open Source Projects

cli authoring must receive the command grammar — flags are parsed facts, not things the model guesses Open

truecourse-ai/truecourse#869 · 1 assignee · Opened Aug 5, 2026

one scenario per flow holds testable claims hostage — a flow must author as many tests as its claims require Open

truecourse-ai/truecourse#868 · 1 comment · 1 assignee · Opened Aug 5, 2026

externals detector: no detection source for spawned binaries — programs the code executes (claude, dotnet, docker) are invisible to the externals model Open

truecourse-ai/truecourse#867 · 1 assignee · Opened Aug 5, 2026

externals detector: aiServices SDK registry predates the Vercel AI SDK — @ai-sdk/* imports are invisible, so Anthropic goes undetected Open

truecourse-ai/truecourse#866 · 1 assignee · Opened Aug 5, 2026

surface matching: stdio protocols are not HTTP APIs — a flow must only match the api surface when an actual HTTP server exists Open

truecourse-ai/truecourse#865 · 1 assignee · Opened Aug 5, 2026

externals detector: the repo's own product domain becomes a detected external service — self-reference must be filtered Open

truecourse-ai/truecourse#864 · 1 assignee · Opened Aug 5, 2026

fidelity/triage LLM calls: 120s budget is mis-sized and a lost call loses its flow attribution Open

truecourse-ai/truecourse#863 · 1 assignee · Opened Aug 5, 2026

cli driver: daemon/service flows need boot-style lifecycle steps — a service flow must be testable, not an eternal infra error Open

truecourse-ai/truecourse#862 · 1 assignee · Opened Aug 5, 2026

guard generate leaves self-heal work silently pending — the retry loop must complete in-run, never depend on an unprompted second run Open

truecourse-ai/truecourse#861 · 4 comments · 1 assignee · Opened Aug 5, 2026

Coarse generate stages freeze mid-call — no liveness heartbeat, no unit counters for few-large-call stages Open

truecourse-ai/truecourse#857 · 3 comments · 1 assignee · Opened Aug 4, 2026

guard run streams every drift as multi-line log output mid-run — results belong to the report, live output to the counter Open

truecourse-ai/truecourse#856 · 1 assignee · Opened Aug 4, 2026

Cold-run guard generate estimate overshoots by ~120× (claim-derived flow bound) Open

truecourse-ai/truecourse#854 · 1 assignee · Opened Aug 4, 2026

Relevance filter keeps repo-internal non-spec docs (docs/findings, test fixtures) — 'ours' is answered, 'is it spec material' is not Open

truecourse-ai/truecourse#851 · 1 comment · 1 assignee · Opened Aug 4, 2026

guard setup proposes a CLI-only recipe even when the engine derives hundreds of API journeys Open

truecourse-ai/truecourse#850 · 1 assignee · Opened Aug 4, 2026

A flow should only show the drivers it belongs to — never 'no code path does this' for every other driver Open

truecourse-ai/truecourse#849 · 1 assignee · Opened Aug 4, 2026

External APIs: 22 'detected but not declared' rows on first setup — inventory shown before relevance is known Open

truecourse-ai/truecourse#844 · 1 comment · 1 assignee · Opened Aug 4, 2026

guard generate aborts on library-only packages with no `bin`/`main` — discards a coverage report the pipeline can already produce Open

Difficulty 4/5 3-5 days Newbie friendliness 52/100

truecourse-ai/truecourse#839 · Opened Jul 31, 2026

guard generate: authoring stage fails 100% on API transport — array-rooted schema vs OpenAI json_object mode (0.7.4-next.2) Open

Difficulty 3/5 1-2 days Newbie friendliness 68/100

truecourse-ai/truecourse#838 · Opened Jul 31, 2026

API mode: OpenAI-family providers reject every stage schema — guard generate and spec scan produce nothing Open

Difficulty 3/5 1-2 days Newbie friendliness 64/100

truecourse-ai/truecourse#836 · 1 comment · Opened Jul 29, 2026

Affected-only test execution — domain split works, selection needs a fail-open design Open

Difficulty 4/5 3-5 days Newbie friendliness 55/100

truecourse-ai/truecourse#825 · Opened Jul 22, 2026

Retire the CLAUDE_CODE_* env namespace Open

Difficulty 5/5 Over a week Newbie friendliness 35/100

truecourse-ai/truecourse#822 · Opened Jul 20, 2026

Nonzero claude exit discards stdout/stderr — errors surface as blank 'claude exited 1:' Open

Difficulty 3/5 1-2 days Newbie friendliness 64/100

truecourse-ai/truecourse#810 · Opened Jul 17, 2026

Empty corpus reads as success: scan says 'nothing changed' and generate says 'every section already guarded' over zero docs Open

truecourse-ai/truecourse#807 · 1 assignee · Opened Jul 17, 2026

spec scan reads only .md — rst/asciidoc spec corpora produce an empty universe Open

truecourse-ai/truecourse#806 · 1 assignee · Opened Jul 17, 2026

LLM call guards: judge progress (TTFT + stall), not wall-clock duration Open

Difficulty 4/5 3-5 days Newbie friendliness 52/100

truecourse-ai/truecourse#805 · 1 comment · Opened Jul 17, 2026

bug: analyze ignores the configured LLM violation model Open

truecourse-ai/truecourse#799 · 1 comment · 1 assignee · Opened Jul 16, 2026

`analyze`: no resume / partial re-run — a rate-limit near the end forces a full ~2.8M-token re-run Open

truecourse-ai/truecourse#791 · 21 comments · 1 assignee · Opened Jul 15, 2026

contracts generate fails on Windows: spawn ENAMETOOLONG (--append-system-prompt exceeds command-line limit) Open

Difficulty 3/5 1-2 days Newbie friendliness 68/100

truecourse-ai/truecourse#660 · Opened Jun 25, 2026

Spec tab "Diff" mode silently drops in-place claim edits (server diffByKey detects only added/removed keys, never "changed") Open

Difficulty 3/5 1-2 days Newbie friendliness 78/100

truecourse-ai/truecourse#594 · Opened Jun 18, 2026

analyze crashes with "Invalid string length" on repos that vendor minified JS Open

Difficulty 4/5 3-5 days Newbie friendliness 45/100

truecourse-ai/truecourse#591 · 1 comment · Opened Jun 17, 2026

verify hard-fails when no contracts exist (generate → verify pipeline breaks) Open

Difficulty 3/5 1-2 days Newbie friendliness 62/100

truecourse-ai/truecourse#590 · Opened Jun 17, 2026

verify: stash sweeps away untracked .truecourse/contracts/, causing ENOENT right after a successful contracts generate Open

Difficulty 4/5 3-5 days Newbie friendliness 55/100

truecourse-ai/truecourse#542 · Opened Jun 8, 2026

`contracts generate` exits 1 with "No contracts were written" on soft validation issues (unresolved cross-references) — but contracts were actually written Open

Difficulty 3/5 1-2 days Newbie friendliness 74/100

truecourse-ai/truecourse#492 · Opened Jun 5, 2026

Llm model provider needed for vscode skills Open

Difficulty 5/5 Over a week Newbie friendliness 35/100

truecourse-ai/truecourse#50 · Opened Apr 18, 2026

Default port reserved by docker Open

Difficulty 3/5 1-2 days Newbie friendliness 35/100

truecourse-ai/truecourse#44 · Opened Apr 17, 2026

Add to GitHub copilot skills Open

Difficulty 3/5 1-2 days Newbie friendliness 35/100

truecourse-ai/truecourse#42 · Opened Apr 17, 2026

Add ability to create custom rules and enable/disable rules Open

good first issue

Difficulty 5/5 Over a week Newbie friendliness 35/100

truecourse-ai/truecourse#9 · 3 comments · Opened Mar 25, 2026

Add dismiss violations functionality Open

good first issue

Difficulty 5/5 Over a week Newbie friendliness 28/100

truecourse-ai/truecourse#8 · 1 comment · Opened Mar 25, 2026

truecourse-ai

Get new issues in your inbox